Description
Scrabble is a commonly played word game in which players take turns forming words using a set of seven letter tiles and placing them onto a grid, following placement rules similar to a crossword puzzle. Points are awarded based on the sum of values of individual letters, along with letter and word multipliers positioned regularly throughout the board. While turns are not usually timed, creativity and vocabulary are both important to achieving a high score. With the increasing popularity of Scrabble-like online games like Words with Friends, various websites like wordfind.com have been created to help players create words from their set of tiles. However, the most challenging (and often most frustrating) part of the game still remains: the player needs to find a spot on the existing board to put their word given Scrabble’s placement rules. Commercial applications rarely take the board state as input; entering the board manually is tedious for the user, but digitizing the board automatically is relatively complex and may be unreliable. This paper presents an image processing algorithm for digitizing Scrabble boards from a single image of the board. The algorithm is evaluated for use with a Scrabble “oracle” backend which takes as input a board state and a set of tiles and suggests the best possible scoring word to the user.
For evaluation of the algorithm, the backend used was a Scrabble oracle written in Scala which determines the best possible word to play given the user’s set of tiles and the current board state.
[1] Gaurav, D. D. and Ramesh, R. (2012). A feature extraction technique based on character geometry for character recognition. CoRR, vol. abs/1202.3884.
[2] Hu, M. (1962). Visual pattern recognition by moment invariants. IRE Trans. Information Theory, vol. 8, no. 2, 179 – 187.
[3] Leung S., Perkins, S., and Rhoades, C. Bananagrams Tile Extraction and Letter Recognition for Rapid Word Suggestion. EE368 Winter 2013-2014
[4] Ohya, J., Shio, A., and Akamatsu, S. (1992). Recognizing characters in scene images. IEEE Transactions on Pattern Analysis and Machine Intelligence, 214-220.
https://github.com/oracle?language=java
Reviews
There are no reviews yet.